Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/zuke-build/zuke/agents-mdgit clone --depth 1 https://github.com/zuke-build/zukeWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.09125 | $0.09125 |
| Opus 5 | $0.04562 | $0.04562 |
| Sonnet 5 | $0.01825 | $0.01825 |
| Haiku 4.5 | $0.00912 | $0.00912 |
Grade A, and why
zuke AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 576 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Guidance for working in this repository. Read this before making changes.
Zuke is a code-first, strongly-typed build automation system for Deno/TypeScript.
This file,
AGENTS.md, is the single source of truth for both humans and agents.CLAUDE.mdis a thin pointer whose entire content is@AGENTS.md, so Claude Code loads this file and there is exactly one copy to maintain.
Using Zuke — the API, without guessing
If you are wiring Zuke into a project, do not guess the API and do not fall
back to Deno.Command/shell. Every operation has a typed wrapper, and the
exact signatures are published — read them:
- Check the package catalogue before writing any command.
llms.txt's## Packagescatalogue (raw: https://raw.githubusercontent.com/zuke-build/zuke/master/llms.txt) and the grouped table inskills/zuke-write-build/references/cheatsheet.mdare the only ways to answer "does a@zuke/<tool>wrapper exist for this CLI?" — per-packagedeno doc jsr:@zuke/<package>can only describe a package whose name you already know; it cannot reveal that a package exists. Reaching forCmdTasks.exec(jsr:@zuke/cmd) or a raw$/Deno.Commandfor a tool that has a@zuke/<tool>package is a bug, not a style choice — it discards typed flags, argv purity, and tool resolution. - One file with the whole typed surface of every package:
llms-full.txtat the repo root.llms.txtis the short index. - A single wrapper on the command line:
deno doc jsr:@zuke/<package>(e.g.deno doc jsr:@zuke/deno). - On each package's JSR page / README: a generated
## APIsection. - The CLI surface — commands, flags, and a build's actual targets: run
zuke --help(ordeno run -A zuke.ts --help). It prints the usage grammar, every reserved command (graph,generate-ci,completions <print|install> <shell>,mcp,resume,runs,cancel,register,doc) and flag, plus the current build's targets — with descriptions and dependencies — and its parameters. So an agent asked to set up or run a build discovers the real command surface live instead of guessing;zuke --listis the targets-only view andzuke --list --jsonemits the whole surface (commands, flags, targets, parameters) as JSON for tools. The written reference isdocs/cli.md, andllms.txtcarries a generated## CLIsection; the same data is available in code via the exporteddescribeCli(build).
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 576 lines · 9,125 tokens per session scan A d2a2be2c40aa
zuke AGENTS.md is an instructions file published in the GitHub repository zuke-build/zuke (33 stars, last pushed 2d ago), licensed MIT. It adds 9,125 tokens to every session, about $0.0456 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
turborepo AGENTS.md
Instructions for vercel/turborepo, covering agents.md, architecture, keeping documentation up to date, pull request guidelines and always run pre-commit/pre-push hooks.
nx CLAUDE.md
Claude Code instructions for nrwl/nx, covering documentation contributions, quick reference, github issue response mode, plan-first mode (default) and immediate implementation mode.
nx AGENTS.md
AGENTS.md instructions for nrwl/nx, covering documentation contributions, quick reference, github issue response mode, plan-first mode (default) and immediate implementation mode.
moon AGENTS.md
Instructions for moonrepo/moon, covering agents.md, repository layout, prerequisites, rust workflow and conventions.
moon CLAUDE.md
Instructions for moonrepo/moon, a project described as: A build system and monorepo management tool for the web ecosystem, written in Rust.
open-bsp-api CLAUDE.md
Claude Code instructions for matiasbattocchia/open-bsp-api, covering claude.md, project overview, local checks must match ci (deno), debugging production edge functions and timestamps.